Molecular features of secondary vascular tissue regeneration after bark girdling in Populus


Author for correspondence:
Xin-Qiang He
Tel: +86 10 62757016


  • Regeneration is a common strategy for plants to repair damage to their tissue after attacks from other organisms or physical assaults. However, how differentiating cells acquire regenerative competence and rebuild the pattern of new tissues remains largely unknown.
  • Using anatomical observation and microarray analysis, we investigated the morphological process and molecular features of secondary vascular tissue regeneration after bark girdling in trees.
  • After bark girdling, new phloem and cambium regenerate from differentiating xylem cells and rebuild secondary vascular tissue pattern within 1 month. Differentiating xylem cells acquire regenerative competence through epigenetic regulation and cell cycle re-entry. The xylem developmental program was blocked, whereas the phloem or cambium program was activated, resulting in the secondary vascular tissue pattern re-establishment. Phytohormones play important roles in vascular tissue regeneration.
  • We propose a model describing the molecular features of secondary vascular tissue regeneration after bark girdling in trees. It provides information for understanding mechanisms of tissue regeneration and pattern formation of the secondary vascular tissues in plants.


Plants and animals are subject to injuries and a variety of assaults during their life span, and regeneration is a common way for them to repair the damaged body parts. In animals, most cells, except for stem cells, lose their regenerative capacity during their differentiation into specific cell types. Recent evidence also shows that the introduction of ‘pluripotency-inducing factors’ (PIFs) may cause a cell to change its ‘fate’ (Yamanaka, 2009; Passier & Mummery, 2010; Vierbuchen et al., 2010). In plants, cells can regenerate new organs and tissues through different pathways; for example, organ or tissue restoration via dedifferentiation and redifferentiation or transdifferentiation, organogenesis via meristem formation in callus, and embryogenic-like ontology via somatic embryogenesis (Cui, 1997; Birnbaum & Alvarado, 2008).

Two basic steps are common in the plant regeneration process: acquisition of competence to regenerate via dedifferentiation, and repatterning of regenerated tissues (Birnbaum & Alvarado, 2008). During the first step, chromatin remodeling and cell cycle re-entry have been suggested to regulate cellular plasticity which is necessary for regeneration (Sustar & Schubiger, 2005; Costa & Shaw, 2007; Grafi et al., 2007; Sena et al., 2009). It is also believed that the specification and regulation mechanisms operating during regeneration are similar to those during normal development because multicellular organisms would not adopt different genetic networks to produce the same structures (Birnbaum & Alvarado, 2008). Furthermore, studies on shoot and root regeneration in Arabidopsis have suggested that the redistribution of phytohormone gradients may induce organ regeneration. Auxin transport is necessary for root regeneration; establishing opposing cytokinin and auxin domains is an important step in shoot meristem regeneration from callus (Xu et al., 2006; Gordon et al., 2007; Grieneisen et al., 2007; Sena et al., 2009). Thanks to the utilization of cell-specific markers in model organisms, significant progress has been made in understanding regeneration mechanisms in plants (Xu et al., 2006; Sena et al., 2009; Sugimoto et al., 2010). However, how differentiating cells acquire regenerative competence and rebuild the pattern of new tissues is still largely unknown, and further investigation in other species and different regeneration systems is necessary.

It is well known that plants can regenerate new bark or vasculature upon wounding (Brown & Sax, 1962; Thompson, 1967; Noel, 1970; Stobbe et al., 2002). When a strip of bark is removed from trees, newly formed periderm and wound cambium develop from the callus on the surface of the secondary xylem, and new phloem is subsequently derived from the wound cambium (Stobbe et al., 2002). However, the response is different in the case of large-scale (as much as 1–2 m) bark girdling in trees. The optimal conditions for secondary vascular tissue (SVT) regeneration after bark girdling on a large scale were first established in Eucommia ulmoides, initially to allow repeated harvesting of bark used in Chinese traditional medicine (Li et al., 1981). The process was subsequently observed in several other trees, including Broussonetia papyrifera, Betula pubescens and Populus tomentosa (Lu et al., 1987; Li & Cui, 1988; Cui et al., 1995; Du et al., 2006; Wang et al., 2009). It is interesting to note that after girdling on a large scale, the newly formed sieve elements and wound cambium are derived from differentiating xylem cells rather than from callus (Li et al., 1981; Pang et al., 2008). Proteomic analysis showed that the changes in gene expression pattern corresponded to the progression of SVT regeneration (Du et al., 2006). However, owing to the limitations of sampling and analyzing methodologies, the molecular mechanisms that enable the secondary xylem cells to switch their fates into multipotent cells and rebuild the SVT pattern are largely unknown.

We show here the morphological and molecular features of the process of cell fate switching in differentiating xylem cells during SVT regeneration after bark girdling in Populus. Because this regeneration system is different from plant shoot and root meristem generation and embryogenic processes, our study provides new information for understanding the mechanisms of tissue regeneration in plants; it is also significant for unveiling the regulation of the SVT pattern formation.

Materials and Methods

Girdling and sampling procedure

Four-year-old trees from a single clonal plantation of healthy Populus tomentosa Carriere located in Renqiu, Hebei province, in northern China (38°50′N, 116°10′E) were girdled in early June 2006. A 1.2-m strip of bark was peeled from trees, starting 0.5 m from the base, and the exposed tissue was wrapped with a transparent plastic sheet. Samples were collected at 0, 4, 6, 9 and 12 d after girdling (DAG). Rectangular blocks (2–2.5 × 4 cm2) consisting of regenerated tissues plus mature xylem were smoothly removed from the stem as described by Uggla & Sundberg (2002). Blocks were then cut into halves. One half of the block was immediately frozen in liquid nitrogen after briefly trimming and then stored at −70°C for tangential cryosectioning, while the other half was divided into small blocks (1.5 × 2 mm2) and fixed in 4% paraformaldehyde in 0.01 M phosphate-buffered saline (PBS; pH 7.2) for microscopic observation. Small strips (3 × 20 mm2) were fixed in 70% ethanol for free hand sectioning.

Histological studies

To detect sieve elements, hand-cut sections were stained with 0.005% aniline blue in 0.15 M K3PO4 (pH 8.2) and then observed under UV light by fluorescence microscopy (Zeiss Axioskop2 plus). Sections of 5 μm thickness were cut from small blocks embedded in LR White resin (Sigma) on a microtome (Leitz 1512, Ernst Leitz GmbH, Wetzlar, Germany), and stained with toluidine blue O and/or aniline blue.

Tangential cryo-sectioning

A series of 20-μm-thick tangential sections was taken for each section sample as described by Uggla & Sundberg (2002) with modification. Regenerated tissues at different stages were isolated by tangential cryosectioning at −24°C with a Leica CM1850 Cryostat (Leica Microsystems Nussloch GmbH, Nussloch, Germany). Transverse sections taken from both ends of the specimen and stained with aniline blue were used to locate the position of tangential sections. Cryosections of regenerated tissues from the same tree at certain stages were collected in a 1.5 ml Eppendorf tube and frozen in liquid nitrogen immediately and then stored at −70°C.

Microarray processing and data analysis

Genomic DNA was isolated from cryosections of differentiating xylem from two trees using a modified CTAB method (Doyle & Doyle, 1990). Total RNA was extracted from cryosections representing five tissues, differentiating xylem, dedifferentiating xylem, regenerated phloem, differentiating regenerated cambium and regenerated cambium, and used for microarray hybridization and following quantitative reverse transcription polymerase chain reaction (qRT-PCR) experiments (see Supporting Information, Methods S1 and Table S5 for details). Three biological replicates were initially used for each tissue; however, failures as a result of low RNA quantities or poor quality led to only two replicates being used for the microarray experiment. The 61K Affymetrix GeneChip® Poplar Genome Array (Affymetrix Inc., Santa Clara, CA, USA) was employed in this study. A genomic DNA-based probe-selection (Xspecies) strategy was adopted to re-annotate probe-sets as described by Hammond et al. (2005, 2006). Sample preparation and array processing were carried out in Nottingham Arabidopsis Stock Centre (NASC, UK) following a one-cycle eukaryotic protocol. RNA signal intensity (.CEL) files were generated using the Microarray Analysis Suite (MAS version 5.0; Affymetrix). Data were further normalized using the Invariant Set Normalization method with dChip (Li & Wong, 2001). Nonlogged mean signal intensity (MSI) values for selected probe-sets across the five samples are presented in Tables S1–S4.

Considering the relative shortage of replicates, we performed model-based conservative estimation by introducing 90% lower confidence bounds (LCBs) of fold change, which is the minimal fold change for a given comparison with at least 90% statistical confidence, to detect the gene differential expression. Standard error (SE) is employed to estimate LCBs and probe-sets with SE < 0.5 were selected for further analysis. A positive LCB indicates that the higher signal was from the first member of the pairwise tissue comparison, while a negative LCB indicates that the higher signal was from the second member of the comparison. A cutoff of LCB ≥ 1.5 in at least one out of the four pairwise comparisons was chosen as the threshold. Moreover, if a gene is expressed (or ‘present’) in one sample group, but not expressed (or ‘absent’) in another group, it was also indicated as ‘differentially expressed’ regardless of the observed expression level. Furthermore, the coefficient of variation (CV = SD/mean signal) was also calculated for each probe-set and used to measure the fluctuation of observed transcripts. Clustering analysis was performed using a DNA-Chip Analyzer (dChip) ( (Li & Wong, 2001). Microarray data and further details of the samples are available through The National Center for Biotechnology Information (NCBI) GEO (GEO submission GSE25309).


Secondary vascular tissue regeneration after bark girdling in Populus

The bark-girdled trees regenerated new SVT within 12 d and formed new bark within 1 month in P. tomentosa (Fig. 1). In the growing season, the vascular cambium produces secondary phloem outwards and secondary xylem inwards. The surface of the exposed trunk was moist and smooth when the bark of P. tomentosa was removed (Fig. 1a). The cambium had been removed with the bark, leaving only differentiating xylem cells on the surface of the trunk (Fig. 2a,b). From 2 to 6 DAG, the girdled area of trunk became protuberant with a soft surface (Fig. 1b). Cross section revealed the formation of callus, resulting from the division of ray cells of differentiating xylem (diX) at 2 DAG (Fig. 2c). Axial diX cells under callus underwent periclinal and transverse divisions, indicating xylem cell dedifferentiation at 4 DAG (Fig. 2d,e,f). From 6 to 9 DAG, the surface of the girdled area turned green (Fig. 1c). Sieve elements were detected by aniline blue staining under fluorescence microscopy at 6 DAG (Fig. 2i), while cambium was not observed at this time (Fig. 2g,h). The sieve elements were present several cell layers under the isodiametric callus cells, arranged between ray cells (Fig. S1a,d) and connected longitudinally to each other (Fig. S1b), which is similar to what was observed in E. ulmoides (Pang et al., 2008). Moreover, typical phloem patterning was also observed tangentially (Fig. S1c). From 9 to 12 DAG, the girdled trunk became greener, dry and hard (Fig. 1d). Continuous and flat regenerated cambial cells were observed in several cell layers under the regenerated phloem at 12 DAG (Fig. 2j–l).

Figure 1.

Morphology of girdled trunk during bark regeneration in Populus tomentosa. The surface of the exposed trunk was moist and smooth when the bark was removed (a). The surface of the girdled trunk became protuberant and soft at 4 d after girdling (DAG), resulting from callus formation (b). It became green at 6 DAG (c), and then became dark green, dry and hard at 12 DAG (d).

Figure 2.

Anatomical observation of secondary vascular tissue (SVT) regeneration after bark girdling in Populus tomentosa. Cambium has been removed with bark during girdling, leaving only differentiating xylem (diX) cells on the surface of the trunk, as shown in the cross section (a) and radial section (b). At 2 d after girdling (DAG), ray cells of diX started to divide and form callus cells (Cal) as shown in the cross section (c). At 4 DAG, more callus cells covered the wounded surface, and axial diX cells under callus underwent periclinal and transverse divisions; dedifferentiating xylem cells (deX) were shown in the cross section (d) and radial section (e). No sieve elements can be detected in deX at this time by aniline blue staining under fluorescence microscopy (f). At 6 DAG, sieve elements were detected by aniline blue staining (i), while cambium was not observed at this time in both cross section (g) and radial section (h). Some of deX under regenerated phloem (rPh) will differentiate into regenerated cambium (diC). At 12 DAG, continuous and flat regenerated cambium (rC) was observed in several cell layers in both cross section (j) and radial section (k) under rPh; it was also shown in an aniline blue-stained radial section (l). Arrowheads show sieve elements. Xy, xylem cells. Bars, 100 μm.

Based upon our morphological observation of twice-performed girdling experiments, we divided the regeneration process into four stages: initially (Stage 0), differentiating xylem cells are left on the surface of the trunk after bark girdling (Figs 1a, 2a, b); Stage I, from 2 to 6 DAG, xylem cells dedifferentiate (Figs 1b, 2c–f); stage II, from 6 to 9 DAG, sieve element formation (Figs 1c, 2g–i); stage III, from 9 to 12 DAG, wound cambium formation (Figs 1d, 2j–l).

Transcriptome profiling indicates global changes of gene expression during SVT regeneration

For transcriptome profiling analysis, a series of 20-μm-thick tangential cryosections were collected at each stage of regeneration, as illustrated in Fig. 3(a). Five tissues, differentiating xylem (diX, stage 0), dedifferentiating xylem (deX, stage I), regenerated phloem (rPh, stage II), differentiating regenerated cambium (diC, stage II) and regenerated cambium (rC, stage III) were sampled for Affymetrix poplar whole-genome array hybridization (Fig. 3a).

Figure 3.

Illustration of sampling procedure and differentially expressed gene (DEG) distribution and quantitative reverse transcription polymerase chain reaction (qRT-PCR) validation of microarray data. (a) Schematic illustration of sampling procedure for microarray and qRT-PCR analysis. A series of 20-μm-thick tangential cryosections was taken for each sample: c. 60- to 80-μm-thick differentiating xylem cells (diX) at stage 0, 40- to 60-μm-thick dedifferentiating xylem (deX) at stage I, regenerated phloem (rPh) and differentiating cambium (diC) at stage II and regenerated cambium (rC) at stage III, as indicated by arrows. Bars with different patterning indicate various tissues (mx, mature xylem). (b) Illustration of the regeneration of phloem and cambium and four pairwise sample comparisons. (c) Numbers of DEGs identified to be up-regulated (striped bars) or down-regulated (filled bars) across distinct regeneration stages. Transcripts with lower confidence bounds (LCBs) ≥ 1.5 in at least one of the four comparisons were flagged as differentially expressed. (d) Comparison between the 90% LCB of fold change reported by microarray and fold change (FC) obtained by qRT-PCR. Genes with LCB > 2, close to 2, 1.5, 1 were chosen for validation. Data were obtained from 28 probe-sets across four comparisons (as shown in Fig. 1b). Log2FC for each comparison is plotted against log2LCB. The intercept of the linear regression line was set to pass through the origin.

Four pairwise comparisons, (a)deX/diX, (b)rPh/deX, (c)diC/deX and (d)rC/diC, were performed (Fig. 3b). Transcripts with 90% LCBs ≥ 1.5 across at least one of the four comparisons were chosen as differentially expressed genes (DEGs). Extensive changes in gene expression were observed across all stages of regeneration (Fig. 3c). SYBR qRT-PCR of selected transcripts with different LCB values suggested a good correlation between qRT-PCR and microarray results (Fig. 3d). The large numbers of up- and down-regulated transcripts from diX to deX indicated dramatic transcriptome changes during the process of xylem dedifferentiation. On the other hand, much smaller changes in gene expression were observed from diC to rC, implying similar molecular features of the two samples (Fig. 3c). DEGs were then classified based on gene ontology (GO); the numbers of up- or down-regulated genes in each GO category are shown in Table 1 and Fig. S2. Genes associated with transcription, transport, stress response and response to hormone stimulus were strongly regulated during regeneration (Fig. S2). Based on the GO annotation, it can be seen that genes related to the chloroplast, plasma membrane, nucleus and endomembrane system were enriched (Fig. S2). Most of the GO biological process and cellular component categories had similar numbers of genes that were up- or down-regulated (Table 1 and Fig. S2). However, in early regeneration (dedifferentiation, deX/diX), genes related to cell cycle, DNA methylation and chromatin regulation were dominantly up-regulated while genes related to cytoskeleton and cell wall biogenesis and Golgi apparatus were strongly down-regulated (Fig. S2). During phloem and cambium regeneration (rPh/deX and diC/deX), there was a preponderance of up-regulated genes related to transcription (Table 1 and Fig. S2). In addition, genes involved in transport, stress and hormone response were dramatically altered during phloem reformation (Fig. S2).

Table 1.   Gene ontology (GO) classification of differentially expressed genes during second vascular tissue regeneration
GO categoryFunction category(a) deX/diX(b) rPh/deX(c) diC/deX(d) rC/diC
  1. deX, dedifferentiating xylem; diX, differentiating xylem; diC, differentiating regenerated cambium; rC, regenerated cambium; rPh, regenerated phloem.

  2. The number of up- or down-regulated genes under each GO category for each sample pairwise comparison is shown.

Biological processTranscription23526818691146521717
Response to stress1608410951744293
Response to hormone stimulus11010267444531102
Signal transduction41772715311132
Cell cycle49933731403
DNA methylation and chromatin assembly or disassembly3766227101
Amino acid metabolic process272978520
Programmed cell death and apoptosis151833321112
Ion homeostasis61411111200
Cell wall biogenesis334542010
Cellular componentChloroplast11624102972721821241523
Plasma membrane5517612982362341542226
Endomembrane system3504872811543131363925
Cell wall21814673905549113
Other intracellular components8174262132611
Endoplasmic reticulum58813730322465
Golgi apparatus2113112206300
Other cellular components12831884102255189237514277

Xylem cells may acquire regenerative competence through epigenetic regulation and cell cycle re-entry

Investigation of the expression patterns of the related genes indicated that epigenetic regulation and cell cycle re-entry were involved in early SVT regeneration. Three DNA methyltransferases, METHYLTRANSFERASE 1 (MET1), DOMAINS REARRANGED METHYLASE 1 and 2 (DRM1/2), and CHROMOMETHYLASE3 (CMT3) (Finnegan & Kovac, 2000), and DEMETER (DME), a DNA glycosylase with DNA demethylation activity (Gehring et al., 2006; Morales-Ruiz et al., 2006), are involved in DNA methylation regulation in plants. Our data showed that one putative CMT3 and two MET transcripts were up-regulated, whereas two candidate DME genes were down-regulated in early regeneration in deX (Fig. 4a). The opposite expression patterns of putative DNA methyltransferases and DME suggested that DNA methylation regulation is involved during early SVT regeneration.

Figure 4.

Transcript expression patterns of epigenetic regulation and cell cycle-related genes during Populus tomentosa secondary vascular tissue (SVT) regeneration. (a) Candidate genes involved in DNA methylation, histone methylation, histone acetylation, chromatin remodeling and polycomb group (PcG) proteins were selected and clustered. (b) Cyclins and cyclin-dependent kinase (CDK) genes with differential expression levels across all samples were identified and clustered. Expression values for each probe-set across all samples were standardized (linearly scaled) to have mean 0 and standard deviation 1 as indicated by red/green-colored squares. The coefficient of variation (CV) for each transcript across samples was used to indicate fluctuation in expression. Complete information for each probe-set in the gene lists can be found in the Supporting Information, Table S1.

In addition to DNA methylation regulation, we found that the expression of histone modification enzymes was changed in SVT regeneration. Two histone demethylase families, amine oxidase (AOD) and Jumonji C (JmjC) domain-containing histone demethylase families, have been identified in the Populus genome (Zhou & Ma, 2008). Among these putative histone demethylases, six AOD domain-containing transcripts (PtAOD1B, PtAOD2B, PtAOD7A, PtAOD8B, and another two AOD family genes) and five JmjC domain-containing transcripts (PtKDM3D, PtPKDM11, PtPKDM8A, PtPKDM8B and PtPDKM7E) showed diverse expression patterns during regeneration (Fig. 4a and Table S1).

Two putative histone methyltransferases, HMT1 and HMT3, were also found to have expression peaks at diX and rC, respectively (Fig. 4a). In addition, probe-sets representing putative Populus histone acetyltransferase (HAT) and histone deacetylase (HDAC) genes (Pandey et al., 2002) were identified in our dataset. Four HAT (HAG2, HAG3, HAF2 and HAC4) and five HDAC (HDA6, 8, 14, 19 and one plant-specific HDAC, HDT1) transcripts were differentially expressed. However, they exhibited different expression patterns at early regeneration stages (Fig. 4a).

Among the homologous genes of chromatin remodeling-related proteins (Boyko & Kovalchuk, 2008) identified in our dataset, some of them showed significant changes in transcriptional level. In early regeneration, one group of genes consisting of DDM1 and LHP1 were up-regulated while another group of genes, including putative MBD2 and MBD8, MOM1 and one ISW2-like gene, were down-regulated (Fig. 4a).

Polycomb group (PcG) proteins and the Polycomb repressive complexes PRC1 and PRC2 play important roles during cell fate switch and regeneration (Costa & Shaw, 2007; Birnbaum & Alvarado, 2008). All putative homologs of Arabidopsis PcG genes (Pien & Grossniklaus, 2007) were identified in our dataset, including several PRC2 subunits, such as SWINGER (SWN), FIE and MSI1 (Fig. 4a). SWN was down-regulated in deX and then increased at a later stage in rPh and rC. Conversely, MSI1, MSI2, ICU2 and FIE were up-regulated in deX and diC (Fig. 4a). In our earlier proteomic study, a protein homolog of SWN was found at 6 DAG and then disappeared at 10 DAG (Du et al., 2006). Our present transcript data together with previous proteomic data suggest a role for PcG in xylem cell dedifferentiation during SVT regeneration.

During SVT regeneration, almost all differentially expressed type-A cyclins (CYCAs) and type-B cyclins (CYCBs) exhibited highly similar profiles with a significant peak in expression at early regeneration in deX (Fig. 4b). However, exceptions, such as CYCA2 and type-D cyclins (CYCDs), were observed. Transcripts for CYCA2;2 and CYCA2;4 showed the highest expression in diX and down-regulation during regeneration (Fig. 4b). In Populus, PttCYCA2 expression remains high until very late in xylem development (Schrader et al., 2004). Expression of AtCYCA2 in Arabidopsis is considered to be associated with competence for cell division (Burssens et al., 2000). These data suggest that diX cells retained their developmental plasticity and have the capacity to change into other cell types in certain conditions, such as wounding. Several probe-sets representing Cyclin D3:1 exhibited high expression in diC or rC (Fig. 4b). Cyclin-dependent kinases (CDKs) and CDK-like (CKL) displayed different expression profiles; of these, two plant-specific CDKB genes (CDKB1;2 and CDKB2;1) followed the general expression trend of most CYCA and CYCB genes (Fig. 4b). Similarly, most histone transcripts also exhibited an expression peak in deX (Table S1). In cell cycle re-entry experiments conducted in Arabidopsis, most CYCAs and CYCBs exhibit a distinct peak in early mitosis and are both considered as mitotic cyclins, while CDKB genes exhibit their expression peak at either the G2 phase or the G2/M transition (Menges et al., 2005). Comparing the observed specific expression pattern of mitotic cyclins and CDKB genes during SVT regeneration with the regulatory mechanism found in cell re-entry experiments suggests the activation of cell cycle re-entry of differentiating xylem cells in their early cell fate switching process. This is also consistent with our anatomical findings during early SVT regeneration in both P. tomentosa (Fig. 2) and Eucommia, in which different cell division phases of immature xylem cells have been observed (Pang et al., 2008).

Down-regulation of the xylem-specific program and activation of phloem and cambium developmental programs during SVT regeneration

To address the identity change of the diX cells, we examined reported xylem marker genes, such as cellulose synthase (CesA) genes, xyloglucan endotransglycosylase genes PttXET16C/PtXTH35, programmed cell death (PCD)-associated XYLEM CYSTEINE PEPTIDASE (XCP), lignin biosynthesis-related genes and xylem specific transcription factors (TFs) (Schrader et al., 2004; Zhao et al., 2005; Geisler-Lee et al., 2006; Suzuki et al., 2006). Eight probe-sets representing xylem-specific Populus CesA and CesA-like (Csl) genes (PtCesA18/3-2, PtrCesA3, PtCesA3, PtCesA7, PtCesA4/5, PtCslA1/2) were identified from our dataset and all eight CesA transcripts exhibited high expression in diX but were absent or had very low expression in regenerated phloem and cambium tissues (Fig. 5a and Table S2). The transcript of PtXTH35 and homolog of Arabidopsis XCP2 also show preferential expression in diX (Fig. 5a and Table S2). Populus lignin biosynthesis-related enzymes (Hamberger et al., 2007; Shi et al., 2010), including PAL, C4H, C3H, 4CL (Poptr4CL3L), HCT, CCoAOMT (PttCCoAOMT1), COMT (PttCCoAOMT1 and 2) and CAD (PoptrCAD1 and PoptrCAD4/PtrCAD2/PtCAD) genes, exhibited a similar expression profile with enrichment in diX and a dramatic decline in regenerating tissues, implying the loss of xylem characteristics in cells during regeneration (Fig. 5a and Table S2).

Figure 5.

Down-regulation of xylem development-related genes during Populus tomentosa secondary vascular tissue (SVT) regeneration. (a) Xylem-specific marker genes and lignification-related enzyme genes. (b) Xylem specification-related transcription factors. Complete information for each probe-set can be found in Supporting Information, Table S2.

HD-ZIP III genes have been demonstrated to promote xylem specification (Emery et al., 2003; Carlsbecker & Helariutta, 2005). In Populus, homologs of ATHB-8 (PttHB8), ATHB-9/PHV (PttHB9) and ATHB-15 (PttHB15) exhibit a steep increase in expression on the xylem side of the cambial zone (Schrader et al., 2004). In our study, PttHB8, PttHB9 and PttHB15 were specifically expressed in diX and distinctly down-regulated in regenerating tissues (Fig. 5b and Table S2). In addition, examination of xylem-biased NAC domain TFs and MYB TFs (Wilkins et al., 2009; Zhong & Ye, 2009; Yamaguchi & Demura, 2010) showed six NAC TFs and a group of Populus R2R3-MYB TFs were diX-enriched with dramatic down-regulation during regeneration, as shown in Fig. 5b and Table S2. Taken together, the enrichment of these genes in diX validated the enrichment of these genes in secondary xylem cells; furthermore, the significant decrease in expression of these genes in regenerating tissues indicates the shutting-off of the xylem specification program and thus the loss of xylem cell identity during regeneration.

On the other hand, our data suggested that phloem and cambium gene expression programs were activated during SVT regeneration. Transcription factors ALTERED PHLOEM DEVELOPMENT (APL) and KANADI (KAN) are required for phloem specification and differentiation (Kerstetter et al., 2001; Bonke et al., 2003; Emery et al., 2003). During SVT regeneration, transcripts of Populus APL and PttKAN1, PttKAN2 were accumulated in the phloem formation stage (stage II) and their expression level remained high until the appearance of cambium at stage III (Fig. 6a). Besides, several other groups of TFs were also found highly expressed in rPh, including G2-like MYB, Dof and NAM (NO APICAL MERISTEM) (Fig. 6a and Table S3). MYB-RELATED PROTEIN 1 (MYR1) and Dof genes have been found to be phloem-abundant in various species (Zhao et al., 2005; Le Hir et al., 2008). A recent report also shows that Dof5.6/HCA2 is expressed in phloem and cambium of Arabidopsis inflorescence stems (Guo et al., 2009). For all Dof genes identified in the Populus genome (Yang et al., 2006), six transcripts showed clear enrichment in rPh or/and rC (Fig. 6a). Among them, PtrDOF03/33 was the closest homolog of Arabidopsis Dof5.6. Collectively, our results provided molecular evidence that groups of TFs such as MYB, Dof and NAM contributed to phloem and cambium regeneration.

Figure 6.

Up-regulation of transcription factors involved in phloem specification (a) and cambium regulation (b) during Populus tomentosa secondary vascular tissue (SVT) regeneration. Complete information for each probe-set in the gene lists can be found in Supporting Information, Table S3.

Furthermore, to assess the structure and function of the newly formed sieve elements as transport phloem cells, potential phloem structure and function-associated genes were examined. Genes encoding sieve elements/companion cells specific jacalin-related lectins, sieve element specifically synthesized serpins, sucrose synthase and phloem-specific metabolic enzymes could highlight common features of phloem functions in various plant species (Le Hir et al., 2008; Lin et al., 2009). Clustering analysis showed that these related genes were abundant in rPh (Table S3). As shown in Figs 3, S1, callose, the marker for sieve elements, was detected in stages II and III by aniline blue. Accordingly, transcripts of callose synthase were up-regulated in rPh (Table S3). Transcripts of hormone biosynthesis-related genes, such as GA4/gibberellin 3 beta-hydroxylase and ACC synthase, and stress response genes were also present in rPh, (Table S3). In our earlier in vitro experiments, isotope tracing of twigs with [14C]sucrose showed that newly formed sieve elements are capable of transporting sugar during the bark regeneration process in Eucommia (Pang et al., 2008). Genes encoding sugar transporters, amino acid carriers, H+-transporting ATPase (AHA10), potassium channels as well as RNA-binding proteins showed an increase in expression in rPh (Table S3) and thus provided molecular evidence of transport function recovery of phloem. Together, our data indicate the activation of the phloem program during SVT regeneration.

Current evidence indicates that a WUSCHEL-CLAVATA-like mediated regulatory model is active in cambial meristems, as in the apical meristems (Elo et al., 2009; Zhang et al., 2011). In Arabidopsis, the PXY-CLE41/CLE44 loop maintains the undifferentiated pluripotent status in the cambium (Etchells & Turner, 2010). WOX4, a WUSCHEL-related HOMEOBOX gene, may function in the TDIF/CLE41/CLE44-TDR/PXY signaling pathway to promote procambial/cambial cell proliferation (Hirakawa et al., 2010). In Populus, Schrader et al. (2004) have also proposed the existence of a similar feedback loop in the cambium. PttCLV1, a putative ortholog of CLAVATA1, showed very low expression in diX and deX and displayed significant up-regulation in regenerated cambial and phloem cells (Fig. 6b). Transcripts of CLV2 and several other leucine-rich repeat kinases also exhibited similar expression patterns to PttCLV1 (Table S3). Conversely, the receptor-like kinase PttRLK3 and a WUSCHEL-related homeobox family gene PttHB2 showed preferential expression in diX and down-regulated across regeneration. This was similar to what has been found in the normal cambial region, where PttHB2 shows a steep increase in expression toward the xylem side (Schrader et al., 2004).

PttANT and PttPNH, Populus homologs of AINTEGUMENTA and PINHEAD/ZWILLE, show a distinct expression peak in the cambium zone, which implies their involvement in regulation of cambial cell proliferation (Schrader et al., 2004). Probe-sets for PttANT;1 and PttPNH were identified and they all exhibited an expression peak in regenerated cambium (Fig. 6b, Table S3). Interestingly, the accumulation of PttANT transcripts was also detected as early as in deX. The up-regulation of one PINHEAD gene during cambium regeneration was reported and predicted to function in cambium initiation rather than maintenance (Wang et al., 2009). However, our data revealed that the expression of PttPNH remained high until cambium formation, indicating that PINHEAD might be involved in cambium maintenance as well (Fig. 6b). Populus orthologs of Arabidopsis Class I KNOX genes were highly expressed in the cambial zone and functional analysis suggested that they are likely involved in the cambial cell identity maintenance (Schrader et al., 2004; Groover et al., 2006; Du et al., 2009). During regeneration, KNOX1 and KNOX6 showed an increasing trend in regenerated phloem and cambium (Fig. 6b). In addition, a class II KNOX gene, KNAT3/KNOX3, showed high expression in rPh and diC but lower signals in rC (Fig. 6b). The GRAS family TFs SHORTROOT (SHR) and SCARECROW (SCR) contribute to vascular cell patterning in Arabidopsis roots (Helariutta et al., 2000; Nakajima & Benfey, 2002). During cambium regeneration, two Populus homologs of SHR (PttSHR1 and PttSHR2) exhibited maximum expression in rC, while four putative SCR-like 6 (SCL6) revealed different expression patterns toward rC (Fig. 6b and Table S3). The expression profiling of these TFs indicates their potential contribution in cambium regeneration.

Further expression profile analysis revealed that members of additional TF families, such as R2R3-MYB, WRKY, bHLH, bZIP, Whirly and CONSTANS, were enriched in regenerated phloem and cambium (Table S3).

Phytohormone regulation during SVT regeneration

Phytohormones play crucial roles in regulating tissue or organ regeneration in plants (Xu et al., 2006; Grieneisen et al., 2007; Birnbaum & Alvarado, 2008; Sena et al., 2009). An auxin gradient provides a positional signal in pattern specification during SVT development (Uggla et al., 1996, 1998; Mwange et al., 2005). Redistribution of auxin gradients may induce root regeneration (Xu et al., 2006) and roots failed to regenerate when auxin transport was blocked (Sena et al., 2009). To look into the molecular evidence for auxin redistribution during SVT regeneration, we checked the expression patterns of polar auxin transport (PAT) genes including auxin influx carriers (PttLAX, AUX1-like) and efflux carriers (PttPIN, PIN1-lke) identified in Populus (Schrader et al., 2003). The results showed that most AUX1-like genes were highly expressed in diX; however, PttLAX2 and another AUX1-like genes were also enriched in rC (Fig. 7a). PttLAX3, on the other hand, showed a similar expression pattern to PttPIN1 and a PIN1-like gene with high expression in rC. A very different result was found for PttPIN3, which was enriched in deX (Fig. 7a). In the wood-forming region, PttLAX1 and PttLAX2 are both expressed in secondary cell wall-forming xylem but PttLAX2 is also expressed in the cambial zone. PttLAX3, PttPIN1 and PttPIN2 are expressed in the cambium and dividing xylem mother cells, while PttPIN3 shows highest expression in the cortex layer (Schrader et al., 2003). The correlation between the dynamic expression patterns of PAT genes across SVT regeneration and their spatial expression patterns in the normal wood-forming region implied that the expression changes of PAT genes might contribute to the resetting up of auxin gradients during phloem and cambium recovery. By using high-performance liquid chromatography and immunolocalization techniques, we have previously shown the significant change of auxin concentration and distribution within the girdled areas during bark reconstitution (Mwange et al., 2003).

Figure 7.

Transcript expression patterns of auxin carriers and auxin receptors genes during secondary vascular tissue (SVT) regeneration. Auxin carrier AUX and PIN genes, auxin receptor TIR and auxin-binding protein1 (ABP1) genes were identified and clustered. The complete information for each probe-set in the gene lists can be found in Supporting Information, Table S4. diX, differentiating xylem; deX, dedifferentiating xylem; rPh, regenerated phloem; diC, differentiating regenerated cambium; rC, regenerated cambium; PAT, polar auxin transport; CV, coefficient of variation.

Alongside polar auxin transport, auxin signaling also plays an essential role in tissue regeneration. A list of auxin-responsive genes is involved in Arabidopsis root-tip regeneration (Sena et al., 2009) and one IAA-induced TF is up-regulated during poplar secondary vascular system regeneration (Wang et al., 2009). Homologs of auxin receptors (TIR1) and genes involved in auxin response, including Auxin/Indole-3-Acetic Acid (Aux/IAA) and Auxin Response Factor (ARF), were identified (Table S4). Analysis of TIR1 and auxin-binding protein1 (ABP1) genes demonstrated that putative TIR1 transcripts were up-regulated during SVT regeneration, whereas ABP1 transcript was down-regulated (Fig. 7b). Recent reports indicate that ABP1 is required for the regulation of early auxin-regulated genes and PAT (Robert et al., 2010; Effendi et al., 2011). Our observation might imply different modes of action for the two potential auxin receptors in SVT regeneration. A total of 35 Aux/IAA genes denoted as PoptrIAA and 39 ARF denoted as PoptrARF were discovered in the Populus genome (Kalluri et al., 2007). Twenty-five probe-sets representing 15 PoptrIAA and 18 probe-sets representing 14 PoptrARF were identified from our dataset and they showed differential expression profiles and tissue preference during regeneration (Table S4). The presence of PoptrARF8 in rPh (Table S4) confirmed previous detection of this gene transcript in phloem (Zhao et al., 2005; Le Hir et al., 2008). Several auxin-responsive genes (examined by Sena et al. (2009)) and enzymes putatively involved in Aux/IAA proteolysis were also up-regulated in rPh and rC (Table S4). The regulatory role of auxin during SVT regeneration has also been supported in our experiments carried out in Eucommia, in which application of exogenous IAA on the wound surface accelerates sieve element differentiation (Pang et al., 2008).

In addition to auxin, genes involved in other phytohormone signalling pathways, such as those of cytokinin, GA and ethylene, were also differentially expressed during SVT regeneration (Table S4). Recent studies demonstrate that cytokinins are central hormonal regulators of cambium development (Matsumoto-Kitano et al., 2008; Nieminen et al., 2008; Dettmer et al., 2009). Cytokinins can accelerate sieve tube regeneration and promote callose production around the wound of Coleus internodes (Aloni et al., 1990). Analysis of the expression patterns of cytokinin receptors, type-A and type-B response regulators (RRs) identified in Populus (Ramirez-Carvajal et al., 2008) revealed that transcripts of putative Populus CRE1B and histidine kinase-like (PtHK3a and PtHK3b), four type-A RRs (PtRR1, 2, 7 and 10) and three type-B RRs (PtRR19, 21 and 22) were up-regulated in rPh and/or rC (Table S4). The GA signal transduction protein (spindly), signal response protein GIBBERELLIC ACID INSENSITIVE (GAI) and GA-regulated protein GASA were differentially expressed during regeneration. However, multiple expression patterns for these genes were observed (Table S4). Homologs of ethylene receptor members EIN4, ERT2 and ERS were differentially expressed during regeneration. A number of Populus ERFs (ethylene response factors) and DREBs (dehydration-responsive element binding proteins) were enriched in regenerating tissues (Table S4).


Root and shoot regeneration in plants has been widely studied but with limited attention paid to vascular tissue regeneration. In this study, we report an in situ plant regeneration system, the SVT regeneration after bark girdling in Populus. Unlike most other de novo regeneration system in plants (Sena & Birnbaum, 2010), SVT regeneration does not lead to entirely new individuals, but only the regeneration of missing tissues in the vasculature (Fig. 2). Furthermore, compared with other in situ regeneration systems, such as Arabidopsis root tip cut regeneration, our system includes cell dedifferentiation, transdifferentiation and new tissue pattern rebuilding, which happens over a longer time span, and this provides us with a good opportunity for higher temporal resolution analysis of plant regeneration.

In SVT regeneration after bark girdling, the wound cambium and sieve elements are not derived from callus but from differentiating xylem cells. After girdling, differentiating xylem cells at different positions transdifferentiate to form phloem cells or dedifferentiate to form the wound cambium (Pang et al., 2008). Furthermore, microarray analysis with different regenerated tissues at distinct stages revealed the molecular features of the regeneration process. It is commonly accepted that cells dedifferentiate to form pluripotent cells (as callus) during regeneration. However, recent findings in plants have suggested that organ regeneration does not need complete dedifferentiation to a pluripotent state (Pang et al., 2008; Sena et al., 2009). On the other hand, although callus has long been regarded as dedifferentiated and pluripotent cell mass, recent studies demonstrated that callus possesses root meristem characteristics even when it is derived from shoot tissues (Atta et al., 2009; Sugimoto et al., 2010). Our experiments showed that phloem or cambium regeneration does not need cell dedifferentiation into a callus state during SVT regeneration in Populus.

To address whether the molecular program of early SVT regeneration is similar to organogenesis via callus, we compared our profiling data in early SVT regeneration (deX/diX) with published profiling data of early callus induction (CIM3/CIM0) in Populus (Bao et al., 2009). During early SVT regeneration, 13% of up-regulated genes (649) and 6% of down-regulated genes (294) were common with these callus-expressed genes, suggesting only modest overlap of molecular processes by these two systems. The enriched GO classes of genes that were common and up-regulated include those related to chloroplast, transcription, transport and response to stress, while the largest common down-regulated class was genes related to the plasma membrane (Table S1, Fig. S2). However, different expression patterns of specific groups of genes were observed when the two processes were compared. Among the 223 up-regulated and 353 down-regulated transcription factors during early callus induction in Populus (Bao et al., 2009), only 47 were up-regulated and 80 down-regulated during early SVT regeneration. The regulation of cell cycle genes was different. CYCA2 genes are up-regulated in CIM3 during early callogenesis (Bao et al., 2009) while down-regulated in deX (Fig. 4b). Moreover, most CYCAs and CYCBs were up-regulated in deX, which is not observed in early callus induction (Bao et al., 2009). Auxin and cytokinin signaling-related genes were also differently regulated during early SVT regeneration (Table S4) and early callus induction (Bao et al., 2009). These data provide molecular evidence that SVT regeneration is dissimilar from in vitro callogenesis.

Chromatin remodeling and cell cycle re-entry are necessary for organ regeneration (Costa & Shaw, 2007; Birnbaum & Alvarado, 2008). Histone methylation controls telomere lengthening in Arabidopsis mesophyll cells undergoing dedifferentiation and is necessary for maintaining the dedifferentiating state of cells and cell cycle re-entry (Grafi et al., 2007). During Drosophila leg to wing transdifferentiation, PcG proteins involved in PRC1 are down-regulated in transdifferentiating cells, and the frequency of transdifferentiation is enhanced in PcG mutant flies (Lee et al., 2005). In Arabidopsis, double mutation of two PcG genes CURLY LEAF (CLF) and SWN induces the formation of ectopic callus and somatic embryos (Chanvivattana et al., 2004), which implies that down-regulation of PcG promotes dedifferentiation from differentiated cells. Our results indicate that epigenetic modulation is involved in the early stages of SVT regeneration and differentiating xylem cells may acquire regenerative competence through cell cycle re-entry during SVT regeneration.

Phytohormones are important in the regulation of new tissue reconstruction. During phloem regeneration, both exogenous auxin and cytokinin accelerate sieve element differentiation in different vascular regeneration systems (Aloni et al., 1990; Pang et al., 2008). On the other hand, auxin transport is necessary for root regeneration, and setting up opposing cytokinin and auxin domains is an important step in shoot meristem regeneration from callus (Xu et al., 2006; Gordon et al., 2007; Grieneisen et al., 2007; Sena et al., 2009). Dramatic expression changes of auxin signaling genes as well as the dynamic expression patterns of PAT genes (Fig. 7 and Table S4) indicate that auxin signaling and redistribution play important roles during SVT regeneration. The expression changes of PAT genes might contribute to re-establishing the auxin gradients and thus provide positional signaling for the specification of phloem and cambium. Based on the gene expression patterns of related genes (Table S4), we conclude that cytokinin and ethylene signaling are also activated during SVT regeneration.

Based upon our morphological observation and transcriptome profiling, we proposed a model for SVT regeneration in Populus (Fig. 8). After bark girdling, differentiating xylem cells which remain on the trunk acquire the competence to switch their fate through epigenetic regulation and cell cycle re-entry. Then, the xylem differentiation program is blocked and the cell fates of the remaining differentiating xylem cells at different locations change at different time points after girdling. This is decided by ‘positional information’ which is coordinated by specific signal molecules. When xylem cells change into a competent state, phloem specification-related genes such as APL, KANADI and Dof are activated and hormone distribution is repatterned, resulting in the re-establishment of an expression program for phloem development. In a similar way, genes involved in cambium determination and maintenance are up-regulated and the hormone gradient is repositioned to promote the reformation of cambium. Concurrently, a cell signaling network consisting of transcription factors and hormone signaling transduction pathways is re-established in regenerated tissues. Thus, the lost phloem and cambium tissues are reconstituted from differentiating xylem cells. Our hypothetical model describes the molecular features of SVT regeneration after bark girdling in trees. It also provides information for revealing the mechanisms of tissue regeneration in plants and for understanding pattern formation of the SVTs.

Figure 8.

A proposed model for differentiating xylem cells switch fate into phloem and cambium during secondary vascular tissue (SVT) regeneration after girdling. After bark girdling, differentiating xylem cells remained on the trunk. Chromatin remodeling and cell cycle re-entry can help to reset the genomic state in early regeneration. While the xylem developmental program is blocked, the phloem or cambium program is activated, resulting in their progressive reestablishment. Phytohormones also play an important role in the regulation of SVT regeneration. NAC:NAM, ATAF and CUC; Dof, DNA binding with one finger; CLV, CLAVATA; ANT1, AINTEGUMENTA1. APL, ALTERED PHLOEM DEVELOPMENT; SCR, SCARECROW; SHR, SHORTROOT; TF, transcription factor;


We thank Ykä Helariutta (University of Helsink, Finland), Hiroo Fukuda (The University of Tokyo, Japan) and Jing-Chu Luo (Peking University, China) for the valuable suggestions on this project and critical comments on the manuscript. We thank Sedeer El-Showk (University of Helsink, Finland) for critical reading and editing of the manuscript. This work was supported by the National Natural Science Foundation of China (31070156), the China Ministry of Agriculture Transgenic Breeding Projects (no. 2009ZX08009-095B) and the China Ministry of Science and Technology 863 Program (no. 2006AA02Z334; 2007AA02Z165).